Epidemiologic studies

Linet MS, Hatch EE, Kleinerman RA, et al. Residential exposure to magnetic fields and acute lymphoblastic leukemia in children. New England J Med 1997;337:l-7.

McBride ML, Gallagher RP, Theriault G, et al. Power-frequency electric and magnetic fields and risk of childhood leukemia in Canada. Am J Epidemiology 1999:149:831-842.

Green LM, Miller AB, Agnew DA, et al. Childhood leukemia and personal monitoring of residential exposures to electric and magnetic fields in Ontario, Canada. Cancer Causes & Control 1999:10:233-243.

Day N, et al. Exposure to power-frequency magnetic fields and the risk of childhood cancer. The Lancet 1999:354:1925-1931

These four retrospective case/control studies of the association between childhood cancer and magnetic fields were begun around 1990 and have been published in the last several years. They were each intended to employ better exposure assessment methodology than earlier studies were felt to have used -- mainly by averaging exposures at different locations and/or times. But the rationale for such exposure averaging has never been clearly spelled out or justified. Briefly, the findings were as follows:

The first study above shows a slight trend in its results toward a modest risk at higher fields -- a result too weak to allow interpretation by itself.

The second study, on the other hand, is so strongly negative in its findings that it actually indicates a statistically significant protective effect against cancer for moderately elevated fields (but not significantly protective for still higher fields). This bizarre finding casts some doubt on the study's methodology in general.

The third study is the exact reverse, having indications of higher risk at more moderately elevated fields than other studies have indicated.

(These two Canadian studies -- having such flamboyant and such disparate results, have raised questions about the unusual recording meter employed in both, and the unique procedure of having case and control children themselves wear the meter after diagnosis and treatment of the cases. We have addressed these questions in: "Leeper E. and Wertheimer N. Potential motion related bias in the worn dosimeter measurements of two childhood leukemia studies, Bioelectromagnetics 2002:23:390-397.")

The fourth study showed little trend of any kind in its findings. However, its exposure assessment methodology is quite different from that of the earlier work (including its use of averaging), so the failure to find a risk increment may simply be due to its asking a different question. In addition, the study population included only a very sparse number of those with the high-field exposures that have been implicated by previous studies. But for leukemia those few high-field exposures were found somewhat more often for cases than controls (five cases against three controls above 4 milligauss). This dearth of high-field subjects may reflect a difference In British electrical distribution, but almost certainly was exacerbated by their use of exposure averaging (which will tend to "pull down" many of the already-rare high exposure levels measured). This is the largest study done to date, though certain short-cuts were employed.

Interpreting the study results -- pooled analyses

So how can a concerned lay person cut through the confusing and contradictory array of scientific findings?

Scientists believe in what is being done and its ability to arrive eventually at a consensus understanding, but we aren't there yet. We like to say, "more work is needed," and we apologize if that sounds like we're running a society for the preservation of jobs for scientists. We are truly sorry not to be able to give the immediate, firm, confident, unanimous answers that people would like; but this is a difficult area.

One approach that has been taken to the uncertainty of the individual results is to pool the data from various epidemiologic studies to create one large "meta-analysis." This approach has several things going for it:

(1) A problem that has plagued interpreting the work in this area is that for the most part the strongest findings have occurred at field levels (above 3 or 4 milligauss) seen at only a small percentage of houses (except if measured at localized "hot spots" in other houses). By combining studies, the effective "sample size" of highly exposed subjects is improved.

(2) An individual study may have certain flaws in its procedures (usually ones that are unsuspected or are unavoidable), so it helps to combine different studies that may at least have different flaws at different places and perhaps in different directions.

(3) If different studies have used different ways of evaluating magnetic fields, it may be possible (often with the cooperation of the original researchers) to extract those portions of each study's data where the field evaluation procedures used were similar to those of the other studies being considered.

It should be said that this kind of meta-analysis is itself difficult, and comes with its own pitfalls. But it does provide an important way of summarizing and combining the work from a number of published studies.

The two pooled analyses listed below attempted to combine comparable results from various published epidemiologic studies of childhood cancer and magnetic fields -- using, as much as possible, similar measurements taken at similar locations and times, though the individual studies may have branched out into various unique analyses of their own (including complex protocols of field averaging for each subject).

Both analyses found, among the various studies, a fairly consistent indication of tendency toward an elevated childhood cancer risk at residential field levels above about 3 or 4 milligauss. In most of the separate studies there had been an excess of cases (vs. controls) seen at such levels, but too few subjects in each study at those levels to draw statistically significant conclusions. However, for the larger group of subjects in the combined data, that high-field-level excess of cases was significant.

Greenland S, Sheppard AR, Kaune WT, Poole C, Kelsh MA. A pooled analysis of magnetic fields, wire codes, and childhood leukemia. Epidemiology 2000; ll:624-634.

Ahlbohm A, Day N, Feychting M, Roman E, Skiner J, Dockerty J, Linet M, McBrice M, Michaelis J, Olsen JH, Tynes T, Verkasallo PK. A pooled analysis of magnetic fields and childhood leukemia. British J Cancer 2000; 83:692-698.

Expert reviews

There have been several reviews of past work in which a group of scientists expert in the area and holding various views was convened to evaluate the scientific findings to date (including addressing the difficult questions of the quality and merits of the work done). The first listed below was the largest undertaking. Each of these groups, with more or less internal unanimity, found that the evidence indicated magnetic fields have a possible link to the risk of leukemia In children:

Assessment of Health Effects from Exposure to Power-Line Frequency Electric and Magnetic Fields. National Institute of Environmental Health Sciences, NIH Publication No. 98-3981 (1998).

Advisory committee report to the U.K. National Radiological Protection Board (2001).

Report of the working panel assembled by the International Agency for Research on Cancer (2001).

Over earlier years, a number of other reviews of the scientific evidence have been published by groups or individuals -- in some cases putting what can only be described as a spin on their descriptions.

Note that despite what has been said, there is laboratory work supporting the possibility of a cancer effect, although it too has its ambiguities and incompleteness -- and although one of the prominent (but perhaps abrasive) players in that area has been successfully pilloried for his pains.

In general, we seem to be muddling through to some kind of consensus that there is indeed something going on here, even if we are a long way from understanding exactly how much, where, and how it happens.

Ed Leeper
6112 Fourmile Canyon
Boulder, CO 80302

303-442-3773